1 Introduction↩︎

Ultrafast pump–probe spectroscopy has become a key technique for exploring nonequilibrium dynamics in high-temperature superconductors [1], particularly the cuprates, where intertwined orders such as superconductivity, charge-density waves, and pseudogap states compete and coexist [2]. In these materials, the photoinduced response is typically strongly dependent on the photon energy of the pump pulse, which determines whether collective or single-particle excitations dominate the transient dynamics. At the lowest photon energies, in the THz range (\(\sim\)​1–30 meV), ultrafast pulses can resonantly drive collective modes such as the Josephson plasma resonance [3], infrared-active phonons [4], or the amplitude (Higgs) mode of the superconducting order parameter [5]. Such excitations have enabled the observation of transient superconducting-like states even above the equilibrium transition temperature, including optically induced interlayer coherence in underdoped YBa\(_2\)Cu\(_3\)O\(_{6+x}\) [6], non-equilibrium superconductivity driven by intense far-infrared pulses [7], and wavelength-dependent enhancement of superconducting correlations in stripe-ordered cuprates [8]. In the mid-infrared (50–200 meV), resonant excitation of lattice vibrations — such as Cu–O stretching modes — induces strong electron–phonon coupling and nonlinear phononic effects [9], [10], producing transient structural distortions that can enhance or even induce superconductivity [11]. At higher photon energies, near-infrared (near-IR, 0.5–1.5 eV) and visible (1.5–3 eV) pulses excite interband and charge-transfer transitions across the Mott or Hubbard gaps [1], [12], injecting high-energy quasiparticles that disrupt the superconducting condensate. These processes enable time-resolved studies of quasiparticle recombination, pseudogap collapse, and charge-order melting [13]–[16]. Despite the differences in excitation channels, the subsequent relaxation pathways often converge toward similar dynamics characterized by quasiparticle thermalization and gap recovery.

Extreme-ultraviolet (EUV) and soft X-ray photons (\(>\)​10 eV), which access deeper core-level transitions and provide element-specific insight into band structure and many-body interactions, have traditionally been used only as probe pulses in time-resolved ARPES [17]–[20]. Tabletop high-harmonic generation (HHG) sources can deliver ultrashort EUV pulses [21], [22], but their typical fluences (on the order of a few μJ/cm\(^2\)) are too low to significantly perturb low-energy collective modes in bulk superconductors. As a result, no HHG-based EUV pumping in superconductors has yet been demonstrated, and most EUV applications remain probe-only. It was only with the advent of ultrashort, high-fluence free-electron laser (FEL) sources that EUV-to-soft X-ray light could be deployed as an effective pump, enabling direct access to nonequilibrium regimes unreachable by tabletop systems.

Here, we report the first use of ultrashort EUV pulses as a pump to drive ultrafast dynamics in the cuprate superconductor (Bi2212), probed with a 1.5 eV near-IR beam. Our results show that electronic excitations ranging from near-infrared charge-transfer to extreme-ultraviolet core-level transitions lead to convergent ultrafast dynamics governed primarily by the absorbed energy density. This establishes EUV free-electron laser spectroscopy as a powerful frontier tool for quantum materials research. By demonstrating that macroscopically similar dynamics emerge from distinct microscopic electronic pathways, the work broadens the scope of pump–probe experiments on correlated matter. Extending this approach to soft X-ray and attosecond FEL sources will allow direct access to the intrinsic timescales of electronic correlations, positioning FEL-based ultrafast spectroscopy as a paradigm for nonequilibrium studies of complex quantum systems.
The measurements were performed at the TIMER endstation of the FERMI-FEL facility, where the operational conditions provide pulses with excitation fluence sufficient to perturb the superconducting condensate. First, EUV-pump/optical-probe experiments were conducted at room temperature by scanning the photon energy of the EUV pump pulses across several core-level resonances, and by varying the FEL fluence. Then, in order to establish the conditions under which extreme ultraviolet excitation could transiently suppress or quench the superconducting state, the experiment was repeated in the superconducting phase, at  30 K, using a 77 eV pump. The fluence dependence of the EUV-induced dynamics differs above and below the superconducting transition temperature \(T_c\) and, remarkably, in both regimes closely resembles that observed with conventional near-IR excitation, revealing a universal non-equilibrium response that spans a wide range of excitation energies. This universality opens new possibilities for high-energy control of correlated states and extends the reach of pump–probe spectroscopy deep into the extreme-ultraviolet domain.

2 Experimental section↩︎

The experiment (schematic in Fig. S1, Supplementary Material) was performed at the TIMER endstation of the FERMI FEL (Elettra Sincrotrone Trieste, Italy). The photon energy of the seeded FEL pulse, \(E_{\mathrm{EUV}}\), used as the pump, was tuned between 23.5 and 177.1 eV by operating both the first and second stages of the FERMI FEL2 source [23]. The excitation energies were selected to target several core-level resonances; specifically, as summarized in Table 1, \(E_{\mathrm{EUV}}\) was tuned across the Cu-3p absorption edge (70.5, 72.1, 75.2, 77.0, and 80 eV), to the O-2s edge (23.5 eV) and the Bi-4f edge (158 eV), as well as off-resonance at 177.1 eV. The pulse duration was 40–60 fs (FWHM), with higher photon energies corresponding to shorter durations [24], and the repetition rate was 25 Hz. The EUV beam was focused onto the sample to a \(350 \times 250~\mu\)m\(^2\) full-width-at-half-maximum (FWHM) spot using a toroidal mirror. A near-IR probe pulse (\(E_{\mathrm{probe}} = 1.58\) eV) was focused to a \(250 \times 250~\mu\)m\(^2\) (FWHM) spot and delayed relative to the pump by a time \(\Delta t\). The probe pulse had a 100 fs duration, a 50 Hz repetition rate, and intersected the FEL beam at a \(13.8 °\) crossing angle. The probe light reflected from the sample was collected by the same focusing lens, separated from the incident probe by a 50:50 beamsplitter, and imaged onto a CCD camera. For each delay \(\Delta t\), 1000 single-shot images were recorded. Because the pump repetition rate was half that of the probe, half of the images corresponded to the pump-on condition and half to pump-off. For each image, the CCD counts were first integrated over a region of interest corresponding to the probe spot on the camera. The transient reflectivity change, \(\Delta R/R\), was then calculated as the normalized difference between the averaged pump-on and pump-off signals (\({\Delta R}/{R} = {(\langle \text{pump-on} \rangle - \langle \text{pump-off} \rangle)}/{\langle \text{pump-off} \rangle}\)).

Measurements were carried out on high-quality underdoped (Bi2212) single crystals (\(\delta = 0.1\)), with a superconducting transition temperature \(T_c = 70\) K, as determined by SQUID magnetometry. The sample temperature was controlled with an open-loop liquid-helium cryostat.

In pump–probe experiments combining EUV excitation with near-IR optical probing, the impact of the EUV photon absorption length on the measured response needs to be taken into account. Two key aspects arise from this.

(i) The volume over which the EUV pump pulse energy is absorbed, and hence the magnitude of the material excitation and its spatial profile, strongly depend on the photon energy, due to the energy-dependent EUV attenuation length (\(L_{\mathrm{abs}}^{\mathrm{EUV}}\)). To account for this, we estimate the average excitation energy density absorbed over one attenuation length as \[I_{\mathrm{exc}} = (1-e^{-1})F/L_{\mathrm{abs}^{\mathrm{EUV}}},\] where \(F\) is the incident fluence. In our experiment, the excitation fluence, \(F\), was adjusted between 0.2 and 5 mJ/cm\(^2\) using a gas attenuator and thin-film filters (Si, Zr, or Mg); the corresponding energy density \(I_{\mathrm{exc}}\) ranged from 50 to 1300 J/cm\(^3\). We note that these values were estimated using the nominal transmission of the beamline and should therefore be regarded as an upper limit. The average lattice temperature increase in the volume within one \(L_{\mathrm{abs}}^{\mathrm{EUV}}\) from the surface can then be estimated as \(\Delta T_{\mathrm{lat}} = I_{\mathrm{exc}} /C_{\mathrm{lat}}\), where \(C_{\mathrm{lat}}\) is the lattice specific heat; at room temperature \(C_{\mathrm{lat}} \simeq 2.27\) Jcm\(^{-3}\)K\(^{-1}\) [25], [26], resulting in local temperature increase up to 500 K; at low temperature (\(C_{\mathrm{lat}} \simeq 0.22\) Jcm\(^{-3}\)K\(^{-1}\) [25]) \(\Delta T_{\mathrm{lat}}\) varies between 200 and 5000 K in the investigated fluence range, suggesting a significant role played by heating in the measured transient reflectivity response. Because \(L_{\mathrm{abs}}^{\mathrm{EUV}}\) depends strongly on \(E_{\mathrm{EUV}}\), not all values of \(I_{\mathrm{exc}}\) were accessible for every photon energy: for the largest \(E_{\mathrm{EUV}}\), having longer \(L_{\mathrm{abs}}^{\mathrm{EUV}}\), the pulse energy is absorbed over a larger volume, resulting in a lower average excitation magnitude.

(ii) The volume probed by the near-IR probe pulse is much larger than the excited volume, because the near-IR absorption length (\(L_{\mathrm{abs}}^{\mathrm{probe}}\)) is up to an order of magnitude greater than that of the EUV pump. Due to the mismatch in optical absorption lengths between pump and probe, the recorded reflectivity change represents a spatial average over both excited and unperturbed regions. In our experiment, the 1.5 eV probe absorption length is \(L_{\mathrm{abs}} ^{\mathrm{probe}}\approx 166\) nm, estimated based on equilibrium optical constants of the material [27]. In contrast, EUV photons are absorbed within just the first few tens of nanometers of the material. Based on CXRO database values [28], the pump penetration length in the \(E_{\mathrm{EUV}}\) range used here varies between 11 and 85 nm (see Table 1). Such values, estimated from atomic scattering factors, align with tabulated X-ray optical data for layered cuprates and oxides, reporting an absorption coefficient \(\alpha_{\mathrm{EUV}} \sim 1.5 \times 10^{6}\;\mathrm{cm}^{-1}\) (absorption length \(\sim\)​7 nm) [29].

Assuming normal incidence and a uniform lateral beam profile, the effective excited volume fraction within the probed region is estimated as \[v = \frac{V_{\mathrm{exc}}}{V_{\mathrm{probe}}} \approx \frac{L_{\mathrm{abs}} ^{\mathrm{EUV}} }{L_{\mathrm{abs}} ^{\mathrm{probe}}}.\] This implies that only a small fraction of the optically probed material, \(v\), is directly excited by the EUV pulse, and that the reflectivity signal from the excited volume is diluted by a factor \(1/v\) when averaged over the full probed depth. This estimate accounts only for pump-induced changes in optical absorption and neglects potential modifications to the real part of the refractive index. It should therefore be regarded as an order-of-magnitude normalization to facilitate comparison across different photon energies, rather than a precise quantitative description. As summarized in Table 1, at 23.5 eV pump energy, only \(~6.5\%\) of the probed volume is excited, whereas the \(V_{\mathrm{exc}}/V_{\mathrm{probe}}\) ratio increases to \(\sim15\%\) around 72-80 eV, and to \(\sim50\%\) at 158-177 eV. This penetration-depth mismatch substantially reduces the amplitude of the measured transient reflectivity and must be accounted for when interpreting the results, which represent spatial averages over the full probed depth.

3 Results↩︎

3.1 Room temperature dynamics↩︎

Room-temperature measurements were first performed to assess the influence of the pump photon energy, to characterize the temporal dynamics associated with EUV excitation, and to benchmark the results against established optical-pump/optical-probe measurements.

Figure 1 (a) shows four representative traces measured at room temperature for four different pump photon energies \(E_{\mathrm{EUV}}\), each taken at an excitation energy density \(I_{\mathrm{exc}} \sim 200\)–\(300\) J/cm\(^3\) (the complete dataset for all employed \(E_{\mathrm{EUV}}\) values is provided in Fig. S2 Supplementary Material). The overall temporal profile remains essentially unchanged when tuning the photon energy across different resonances, aside from variations in the transient reflectivity amplitude. Increasing the excitation intensity likewise leaves the shape of the dynamics largely unaffected (see Fig. S2 Supplementary Material for full fluence-dependent measurements).

To quantify these observations and extract recovery timescales, the room-temperature \(\Delta R/R\) traces are fitted with a two-exponential decay function, \[I_{\mathrm{RT}} (\Delta t) = a_1 e^{-\Delta t / \tau_1} + a_2 e^{-\Delta t / \tau_2} + c,\] multiplied by a Heaviside step function and convoluted with a Gaussian of 120 fs FWHM to account for the pump and probe pulse durations. The dynamics are well reproduced by two components with time constants \(\tau_1\) and \(\tau_2\), which describe the thermalization of excited electrons with different degrees of freedom, including lattice, strongly coupled phonons, spin and bosonic excitations of electronic origin [26].

A summary of fit results is shown in Fig. 2, where the extracted parameters are plotted as a function of \(I_{\mathrm{exc}}\). For all pump photon energies, the fast decay time \(\tau_1\) ranges from 100 fs to 300 fs, increasing with excitation fluence (see Fig. 2a). This fluence-dependent slowing down is consistent with predictions from two-temperature model rate equations, in which stronger excitation produces a larger instantaneous rise in electronic temperature, thereby reducing the recovery rate [1], [30]. The datapoints relative to \(E_{\mathrm{EUV}} = 23.5\) eV (O-2s shallow core level) slightly deviated from the general \(\tau_1\) trend. We note that for this photon energy, resonant with oxygen states, the excitation fluence may be significantly overestimated, since oxidation of beamline filters and mirrors can substantially reduce the actual transmission relative to nominal values. The slow decay time \(\tau_2\) (Fig. 2b) is \(\sim\)​2–5 ps, with large uncertainties preventing the identification of clear trends with either pump photon energy or excitation density.

The amplitude parameters \(a_1\), \(a_2\), and \(c\) increase approximately linearly with \(I_{\mathrm{exc}}\) for all \(E_{\mathrm{EUV}}\) values (see Fig. S3, Supplementary Material). To compare amplitudes across pump photon energies, we need to account for variations in the fraction of the probed volume that is excited, \(v = V_{\mathrm{exc}} / V_{\mathrm{probe}}\). For this reason, in Fig. 2c–e we report the parameters \(a_1\), \(a_2\) and \(c\) normalized by the corresponding value of \(v\) (see Table 1), while the not-normalized values are provided in Fig. S3. This normalization yields comparable amplitude values for all \(E_{\mathrm{EUV}}\) at a given excitation density, indicating that differences in \(\Delta R/R\) amplitude arise primarily from the short EUV absorption length and the resulting variation in excited volume.

These results demonstrate that the room-temperature response is largely independent of the EUV pump photon energy. This universality also extends to lower photon energies: in particular, we compare the EUV-pump–induced dynamics with those obtained under conventional optical excitation at 1.5 eV. The optical-pump trace (adapted from Ref. ), overlaid with the FEL-pump trace at \(E_{\mathrm{EUV}} = 77\) eV in Fig. 1b, is characterized by decay times \(\tau_1 = 190 \pm 40\) fs and \(\tau_2 = 1.3 \pm 0.8\) ps (see Fig. S6 Supplementary Material for the complete set of fit parameters), in excellent agreement with the dynamics measured under FEL excitation.

In addition to the decay dynamics discussed above, the FEL-pump high-temperature data consistently reveal a weak oscillatory component in the \(\sim3-20\) ps range. This oscillation has a period of approximately 6 ps, corresponding to a frequency of about 150 GHz, as extracted from fits reported in the Supplementary Material (see Fig. S4 and S5 Supplementary Material). The oscillatory behavior is generally absent when the transient reflectivity signal is weak—either at low excitation fluence or when using the 23.5 eV pump. Its origin remains unclear: the frequency is too low to correspond to a response of acoustic origin, yet it could be compatible with an ultra-low-frequency Raman-active excitation, as similarly reported in Bi2212 [31]. Interestingly, this feature is observed only under EUV excitation and not in conventional optical-pump measurements, although in the latter case comparable signal amplitudes may not be available.

3.2 Superconducting phase response↩︎

Low-temperature measurements were performed at \(T = 30\) K (\(< T_c\)) with \(E_{\mathrm{EUV}} = 77\) eV to investigate the temporal response to EUV excitation in the superconducting phase, assess whether superconductivity can be partially suppressed or fully quenched, and determine the corresponding critical fluence.

Cooling the sample below \(T_c\) produces a striking change in the transient reflectivity response, whose temporal shape — rather than just its amplitude — becomes strongly dependent on the excitation fluence (Fig. 3a). At low excitation densities (\(I_{\mathrm{EUV}} \lesssim 100\) J/cm\(^3\)), the optical dynamics deviate markedly from the room-temperature behavior: a delayed reflectivity component emerges (indicated by the gray arrow in Fig. 3a), building up over several hundred femtoseconds and peaking at \(\sim\)​1.5 ps at the lowest measured fluence. This slow component, absent in the normal state, is characteristic of quasiparticle recombination and condensate recovery, and is attributed to a boson bottleneck mechanism [1], [32]. As the fluence increases (\(I_{\mathrm{EUV}} =200–1000\) J/cm\(^3\)), this delayed component vanishes, and the signal evolves toward the normal-state profile. At fluences \(I_{\mathrm{EUV}} \geq 200\) J/cm\(^3\), the dynamics are dominated by the same fast decay (100–300 fs, depending on intensity) observed at room temperature. The disappearance of the slow build-up channel at large excitation intensity indicates complete suppression of the superconducting condensate under these conditions. This abrupt crossover underscores the role of excitation density in driving the system across the phase boundary and defines a threshold regime for EUV-induced condensate destruction.

The observed behavior closely mirrors previous optical-pump/optical-probe measurements on cuprates [32], as well as additional near-IR results acquired for comparison (see Fig. S8 Supplementary Material). In these cases, a delayed peak appears at excitation energy density \(I_{\mathrm{opt}}\) up to \(\sim\)​23 J/cm\(^3\) (corresponding to fluence \(F\approx600\) μJ/cm\(^2\)) and shifts to longer delays as the fluence increases (see Fig. S8b Supplementary Material). Based on the data reported in Ref. , for instance, with a near-IR pump excitation intensity of \(\sim\)​10 J/cm\(^3\) (\(F\sim\)​250 \(\mu\)J/cm\(^2\)), the transient reflectivity dynamics resembles those measured with the FEL pump at the lowest EUV fluence (red trace in Fig. 3a), peaking at \(\sim\)​1.5 ps. Figure 3b compares the low-temperature dynamics for the two excitation schemes: EUV pumping at 77 eV and near-IR pumping at 1.5 eV. The optical-pump trace, adapted from Ref. , was taken at \(T = 20\) K and \(F = 285\) \(\mu\)J/cm\(^2\); the FEL-pump trace was recorded at \(T = 30\) K and \(F = 220\) \(\mu\)J/cm\(^2\). Despite the similar nominal fluences, the corresponding absorbed energy densities differ substantially — \(\sim\)​60 J/cm\(^3\) for EUV versus \(\sim\)​11 J/cm\(^3\) for near-IR — due to the greater penetration depth of the near-IR light. This \(\sim\)​6-fold discrepancy reflects the ratio of excited to probed volume (\(1/v \approx 6.8\)), explaining why higher absorbed energy density is required for EUV excitation to achieve dynamics comparable to those obtained with near-IR light.

\(\Delta R/R\) dynamics fits for \(T < T_c\) were performed using the expression \[I_{\mathrm{LT}} (\Delta t) = a_1 e^{-\Delta t / \tau_1} + a_2 e^{-\Delta t / \tau_2} + a_3 e^{-\Delta t / \tau_3} + c,\] multiplied by a Heaviside step function and convoluted with a Gaussian of FWHM 120 fs. The resulting fit parameters are shown in Fig. 4, while the full set of fitted traces is reported in Fig. S7 Supplementary Materials. For excitation intensities \(\leq 100\) J/cm\(^3\), an additional term with negative amplitude (\(a_3\)), compared to the room-temperature fit function, is required to capture the delayed “hump” associated with condensate recovery (Fig. 4c and f). At the lowest excitation intensity, we find \(\tau_3 = 400 \pm 200\) fs. Applying the same fitting procedure to the optical-pump data in Fig. 3b yields \(\tau_3 = 500 \pm 200\) fs (see Fig. S8a Supplementary Materials). The similar order of magnitude of \(\tau_3\) in the two excitation schemes confirms the comparable underlying dynamics, with small differences attributable to the different temperatures and excitation intensities in the respective measurements.

4 Discussion: Possible Excitation Mechanism Above and Below \(T_c\)↩︎

The striking similarity between the transient optical responses induced by EUV (77 eV) and near-infrared (1.5 eV) pulses—both above and below the superconducting transition temperature—suggests that the underlying excitation mechanism is governed more by the absorbed energy density than by the specific nature of the initial photoexcited states. While EUV photons nominally excite deep valence or even core-level electrons, the data indicate that the energy is rapidly redistributed through a cascade of secondary scattering events, leading to a thermalized population of high-energy quasiparticles in the conduction band. This ultrafast energy redistribution appears to funnel the initial excitation into the same low-energy degrees of freedom that are directly accessed via near-IR excitation.

Above \(T_c\), both EUV and near-IR excitation generate a broad continuum of quasiparticles, likely originating from charge-transfer transitions involving O-2p and Cu-3d states. These carriers quickly relax into lower-lying electronic states via electron–electron and electron–phonon scattering. The measured reflectivity dynamics—featuring a fast rise and a multicomponent decay—suggest a rapid establishment of a quasi-thermal electronic distribution, regardless of whether it is seeded by a shallow interband transition or a deep photoionization event.

Below \(T_c\), a distinct delayed component emerges in the reflectivity signal, peaking at \(\sim\)​1.5 ps for both EUV and near-IR pump types at low fluence. This feature is a hallmark of quasiparticle recombination and superconducting gap recovery [32]. Its persistence under EUV excitation indicates that the superconducting condensate is not immediately destroyed, but rather partially depleted and allowed to recover. This observation implies that, despite their high photon energy, EUV pulses can couple to the superconducting order, likely through their ability to generate dense quasiparticle populations that interact with the condensate indirectly. The recombination dynamics then follow similar bottleneck-limited pathways as seen in optical excitation, involving boson-mediated pair recovery processes. Overall, these results support a picture in which the macroscopic ultrafast dynamics—such as quasiparticle relaxation and condensate recovery—are largely independent of the microscopic excitation pathway, provided the deposited energy exceeds the relevant thresholds. This universality suggests that even high-energy EUV photons, typically associated with core-level spectroscopy, can be harnessed to manipulate low-energy quantum states in a controlled manner, provided the excitation density is appropriately calibrated. Such insights open the door to extending control of correlated phases into new spectral and spatial regimes using FEL-based ultrafast techniques.

5 Outlook: Speculative Dynamics under Soft X-ray and Attosecond Excitation in Cuprates↩︎

Extending ultrafast spectroscopy of cuprates to the soft X-ray and attosecond regimes enables access to correlation dynamics on intrinsic timescales [33], [34]. Unlike optical pumping, soft X-rays at the O-1s or Cu-2p edges create localized core holes that decay via Auger or radiative processes within a few femtoseconds, producing hot carriers and structural distortions. These local perturbations can suppress superconductivity non-thermally or disrupt pseudogap correlations on femtosecond scales [35]. Attosecond soft X-ray pulses (\(<300\) as, \(>200\) eV) allow direct tracking of core-hole creation, screening, and charge redistribution [34], [36]. In cuprates, this could reveal valence charge transfer, gap renormalization, and doublon–holon dynamics. Experimental approaches include attosecond pump–probe, streaking, and two-color schemes to disentangle electronic and lattice responses. Furthermore, attosecond soft X-ray experiments promise a revolutionary view into the microscopic choreography of electronic correlations. In cuprates, such tools could directly interrogate how Mott physics, charge transfer, and superconducting order dynamically emerge, collapse, or compete—all before lattice responses even begin.

FEL-based attosecond sources promise sufficient fluence for pump–probe control, while HHG-based sources remain powerful probes in the soft X-ray regime [37]–[41]. Combined with advanced many-body simulations (tDMRG, non-equilibrium DMFT), these approaches could open new routes to probe and manipulate correlated quantum states.

6 Conclusions↩︎

Our experiment establishes a energy-density–driven pathway for electronic excitations, in clear contrast to bosonic driving where resonance-selective dynamics prevail. This underscores the need to distinguish electronic and bosonic routes to nonequilibrium superconductivity in the Bi2212 cuprate superconductor.
Deep UV or X-rays and optical photons excite very different states, yet both lead to the same quasiparticle bottlenecks and condensate recovery, highlighting a fundamental universality of electronic excitations in cuprates.
Using FEL-EUV pulses at 77 eV, we have shown that it is possible to perturb the superconducting state without immediately destroying it, even when the excitation originates from deep-lying electronic levels. Remarkably, the observed transient reflectivity dynamics closely mirror those triggered by conventional 1.5 eV near-IR pumping across both the superconducting and normal phases. This striking similarity suggests a universal photoresponse that is governed more by the absorbed energy density than by the specific nature of the initial excitation pathway.

Below the superconducting transition temperature, both excitation schemes give rise to a delayed reflectivity component indicative of quasiparticle recombination and condensate recovery. The emergence of this feature under EUV pumping demonstrates that the superconducting state retains its coherence even under high-energy excitation, as long as the energy density remains below the destruction threshold. The main difference lies in the amount of absorbed energy required: due to the much shorter EUV penetration depth, approximately an order of magnitude more energy must be deposited in a localized region to produce a comparable effect, leading to inhomogeneous excitation and a diluted probe signal.

Our experiment represents a first step toward clarifying whether the observed dynamics reflect direct coupling to the condensate or are instead mediated by indirect scattering cascades. This question could be further addressed by complementing our data with theoretical modeling, such as non-equilibrium dynamical mean-field theory (NE-DMFT) [[42]][43][[44]][45] or time-dependent density matrix renormalization group (tDMRG) [46] [47] [[48]][49] studies focused on cascade thermalization or multi-step relaxation, where an initial non-thermal distribution of high-energy excitations (e.g., electrons coupled to bosons) relaxes via intermediate scattering channels, before reaching thermal equilibrium.

These findings are paradigmatic for establishing FEL-based EUV excitations as a foundation for next-generation pump–probe studies of quantum materials, enabling access to regimes beyond optical techniques and opening pathways toward soft X-ray and attosecond multidimensional spectroscopies of correlated phases.

References↩︎